Segfaulting programs with GHC 6.4.1

[John Goerzen]

On Fri, Oct 21, 2005 at 03:17:34PM +0200, Krasimir Angelov wrote:
> If I remember exactly you said that your program was working before
> movement to ghc 6.4.1. If that is the case then I doubt that the
> problem is in HSQL. In any case the easy way to catch that is to add
> some trace messages in HSQL. At least traces at the beginning and at
> the end of query, execute and getColValue functions should be enough.
> Note that I have never tried HSQL with the multithreaded RTS.

Let me just give an update on this.

I have continued to add debug information but still cannot get a
reliable crash point.

I have also switched to forkIO in an attempt to narrow down the problem.
It didn't appear to make any difference.  gb175

I've seen some crashes in HSQL code appear to happen before the 244 in
this snippet:

                getColValue pRes tupleIndex countTuples colNumber
(name,sqlType,nullable) f = do
                        ldebug "getcolvalue entered"
                        index <- readMVar tupleIndex
                        ldebug "244"
                        when (index >= countTuples) (throwDyn SqlNoData)
                        ldebug "246"

(so, on the readMVar line.)

Others appeared to occur after the 261 here:

                                        ldebug "259"
                                        case mb_value of
                                                Just v   -> do ldebug "261"
                                                               return (Just v)

I'll let you know if I can find out anything more.